Implications of on-farm segregation for valuable milk characteristics

Milk composition varies between herds and cows within herds, enabling its segregation on farm, rather than during processing, for the manufacture of specific dairy products. Benefits may include increased product yields, reduced processing costs and greater suitability of differentiated milk for the production of high value niche market products. However, costs are also likely to be greater. An integrated spreadsheet model was developed to determine the break-even premium required for a farmer with a seasonal calving herd to be economically no worse off producing segregated than conventional milk. The model incorporated breeding (quantitative and qualitative traits), cow requirements and feeding, transport, and economic sub-models. Cows were segregated within herd and milk composition was altered over time by genetic selection. Four quantitative trait (“white” milk colour) scenarios and two qualitative trait (BB β-lactoglobulin milk) scenarios were considered.The model suggested that “white” milkfat would need to earn 38.4% more at the farm gate than conventional milkfat for the two systems to break even. “White” milk cows produced less than their status quo counterparts due to the reduced selection pressure on production milk traits and this had a considerable impact on the premium, as did the low initial volumes of white milkfat. The difference in production between the B β-lactoglobulin cows and their status quo counterparts was less than for selection on white milkfat only. The high risk to farmers of discontinuing a differentiated milk policy could be moderated by changing the structure of premium payments over time. Hence, processing companies and farmers will need to work together to facilitate the uptake of milk segregation. This research model could be applied by dairy companies and farmers considering milk segregation policies.     

利用改进BP神经网络预测初产奶牛产奶量的方法

针对传统BP算法的缺陷，提出了一种采用L-M训练法的BP神经网络。在此基础上建立了基于改进BP神经网络非线性系统的奶牛305d产奶量预测模型，此模型可以提前215d预测初产奶牛305d产奶量，从而实现提早进行选择，加速奶牛育种工作进程。并通过具体的实验验证了改进BP神经网络预测模型的有效性。     

现代都市型农业背景下生鲜乳价格波动预警研究——以天津市为例

为稳定生鲜乳价格、保障生鲜乳供给,结合现代都市型农业特点,以天津市为例,构建生鲜乳价格波动预警机制。运用季节调整法和Holt双参数指数平滑法对生鲜乳月度价格进行组合预测并进行预警;利用BP神经网络模型,建立生鲜乳年度价格波动预警机制。研究结果表明,生鲜乳月度价格组合预测结果比较理想,2022年1—6月价格基本维持在434~4.38元/kg,生鲜乳价格波动存在比较明显的季节性,每年年末生鲜乳价格波动易出现正向轻警甚至正向重警;对于生鲜乳年度价格波动,BP神经网络模型预测精度较高,2019年测试结果为0.101 5,与2019年实际值0.106 7相近且均显示为无警状态,可以用此模型对生鲜乳年度价格波动进行预警。为稳定生鲜乳价格,应建立科学有效的市场信息收集、分析、发布机制,加强生鲜乳价格波动预警监测,建立并优化生鲜乳价格协商机制。     

凝乳超声衰减模型及计算机模拟

通过分析酶促凝乳硬度变化指数方程以及超声波在凝乳中的衰减特性，提出超声波在凝乳中衰减的指数模型。用阻尼最小二乘法进行拟合，求得模型中的3个特性参数。实验结果表明，该模型能很好地模拟牛奶的凝固过程。与多项式模型相比，指数模型中的参数具有与牛奶凝固特性相关的物理意义，且拟合精度较高。     

Development and evaluation of a pastoral simulation model that predicts dairy cattle performance based on animal genotype and environmental sensitivity information

A dairy cattle simulation model for pastoral systems that considers how dairy cow genotypes respond to different environments is described. The dairy cow is represented by five modules for maintenance, pregnancy, growth, body energy reserves and lactation with the influence of environmental factors on processes included within each module. Feed intake is predicted based on the requirements for maintenance, growth and pregnancy, and the dairy cow’s potential for yields of milk, fat and protein and body fat change in a given environment. The effects of various temporary environmental factors such as cow body condition score, climate, feed quality and the stage of pregnancy are all considered when predicting yields of milk, fat and protein, energy and dry matter intake. The model was evaluated using information from a prior experimental study with 1990s Holstein-Friesian dairy cattle of North American/European or New Zealand origin managed in a pasture-based system in early to peak lactation. The model was able to predict, to a high degree of accuracy, mean values for yields of milk, fat and protein, and concentrations of fat and protein. However for individual cows, feed intake and live weight change were less reliably predicted. The major source of error was a lack of simulated variation, rather than any systematic bias. The major advance of the model is its ability to predict performance from genetic and environmental sensitivity information for particular breeds, and its ability to predict feed intake and yields of milk, fat and protein concurrently.     

Simulating the influence of management decisions on the nutrient balance of dairy farms

In order to evaluate the influence of management decisions on the nutrient balance of dairy farms a simulation model was developed. Three farm systems have been simulated: zero grazing, winter milk and summer milk. From the simulated farm systems the zero grazing farm has in all scenarios the lowest N-surplus. The winter milk farm system has a higher N-surplus than zero grazing but lower than the summer milk farm system. The results further indicate the positive effects of maize feeding in addition to grazing. More maize in the ration is especially good to lower the N-surplus during the grazing period in the summer. The benefits of more maize in the ration decrease when the fertilizer application rates decrease.     

A systems approach to quantify greenhouse gas fluxes from pastoral dairy production as affected by management regime

A model was developed to determine what effect management practices would have on the production of the greenhouse gases (GHG) within pastorally based dairy production systems typical of those practiced in Ireland. The model simulates two levels of GHG emissions, firstly the on-farm GHG emissions of methane, nitrous oxide and carbon dioxide for example from the pastorally spreading of slurry and secondly, off-farm GHG emissions associated with both inputs brought onto the farm to maintain productivity (for example emissions arising from manufacture of concentrate feeds and fertiliser) as well as from indirect GHG emissions associated with nitrate leaching and ammonia. The aim of this work was to allow the development of effective GHG mitigation strategies at the farm level capable of reducing GHG emissions per litre of milk.Greenhouse gas emissions were modelled for nine farming systems differing in the level of concentrate supplementation (376, 810 and 1540 kg per cow per lactation) and genotype for milk production as assessed by their pedigree index (<100, 100–200 and 200–300 kg) of milk production. A three-year study to evaluate the influence of cow genetic potential for milk production and concentrate supplementation level on profitability of pasture-based systems of milk production was used to drive the Moorepark Dairy Systems Model (MDSM). Output from this model then described farm size, feed budgets, animal numbers and farm profitability when annual milk quota was set to 468,000 kg of milk year. Relating GHG emissions to annual milk sales revealed that for these pastorally based systems increasing concentrate usage reduced both on-farm and off-farm emissions, but that increasing the genotype of the dairy cow (i.e., the genetic capacity of the animal to produce milk) will increase both on-farm and off-farm GHG emissions. Lowest GHG emissions per kilogram of milk were achieved for an intermediate genotype type cow fed within a high concentrate system whilst the highest emissions were associated with high genotype cows fed within a low concentrate system. Maximum profitability was obtained when either a high concentrate feeding regime was combined with high genotype cows or where low concentrate systems were fed to low genotype cows.Relating farm profitability to GHG emissions allowed the identification of scenarios where changing from one management systems to another would achieve a simultaneous reduction in GHG emissions whilst improving farm profitability. By implementing this approach of assessing management induced change on both GHG emissions arising from the farm together with farm profitability, individual whole farm GHG mitigation strategies could be developed with a high degree of acceptability to the producer.     

Simulation of the intake and partition of nutrients by the dairy cow: Part II— The yield and composition of milk

This paper is the second in a series describing a computer-based simulation model designed for use as a management control aid for the dairy herd and concerned with the intake and partition of nutrients by the cow. Milk production is represented in the model in two parts; the first estimates the potential milk yield for each 24 h simulation interval and the second is concerned with matching the available metabolites to the requirements associated with the potential yield. In each case, a conceptual scheme for inclusion in the model is outlined in relation to the relevant literature and the manner in which these schemes are represented in the model is described.     

Exploring changes in world ruminant production systems

In the past 30 years world production of ruminant meat and milk has increased by about 40%, while the global area of grassland has increased by only 4%. This is because most of the increase in ruminant meat and milk production has been achieved by increasing the production in mixed and landless production systems and much less so in pastoral systems. Pastoral systems depend almost exclusively on grazing, while mixed and landless systems rely on a mix of concentrates (food crops) and roughage, consisting of grass, fodder crops, crop residues, and other sources of feedstuffs. A model was developed to describe these two aggregated production systems for different world regions, each having typical production characteristics, such as milk production per animal for dairy cattle, and off-take rates and carcass weights for non-dairy cattle, sheep and goats. The energy needed by the animals for the production of meat and milk is calculated on the basis of requirements for maintenance, grazing and labour, pregnancy, and lactation. We implemented the FAO Agriculture Towards 2030 projection for crop and livestock production and assumed that the past trend in the area of grassland will continue in the coming three decades. This assumption implies a rapid intensification of grassland management with a 33% increase in global grass consumption, which will only be possible with increasing fertilizer inputs, use of grass-clover mixtures and improved grassland management.     

A mathematical model for the joint metabolism of nitrogen and energy in cattle

A mathematical model for the joint metabolism of nitrogen and energy has been developed and its utility studied. It takes into account digestion, absorption, anabolism, catabolism, and excretion of nitrogenous and non-nitrogenous nutrients. The model consists of a set of non-linear differential equations which, upon integration, yield predictions for changes in body proper and energy stores and for outputs of milk, heat, methane, and nitrogenous and non-nitrogenous wastes. Key aspects are the modes whereby nitrogenous and non-nitrogenous nutrients combine for body growth and milk production and nitrogenous nutrients split into non-nitrogenous nutrients and nitrogenous wastes.Values for some of the model parameters were obtained from the literature, but many of the important ones had to be estimated from experimental data. Estimation was accomplished by iterative weighted least squares fitting of the model to published results from energy and nitrogen metabolism trials on growing steers and on milking and non-milking cows fed widely-varying protein and energy intakes.The model was tested with metabolism data from trials not used in estimating parameters. Over 70% of the comparisons for daily output of faecal nitrogen and energy, urine nitrogen and energy, milk nitrogen and energy, and methane showed deviations between prediction and observation of 20% or less. These deviations were little greater than the discrepancy between observations for replicate trials, thus establishing a basic soundness of the model.Although fitted and tested with cattle, the model applies in principle to animals generally and man, for predicting outputs from widely-varying nitrogen and energy inputs as conditioned by species, type and physiological state. The model, with some extension, should permit more profitable ration formulation than do extant approaches over a very broad domain of feeding and price situations.     

婴幼儿奶粉中多种掺假物近红外高光谱图像检测方法

奶粉市场是食品掺假行为频发领域,其中婴幼儿配方奶粉价格高,其质量是消费者、生产企业和执法部门关注的重点。近红外高光谱成像（Near infrared-hyperspectral imaging, NIR-HSI）技术结合化学计量学和机器学习算法可以检测奶粉中单一掺假物含量。基于NIR-HSI技术研究了不同品牌婴幼儿奶粉中多掺假物（三聚氰胺、香兰素和淀粉）的定量预测。对基于像素点预处理后的高光谱图像划分感兴趣区域（Region of interest, ROI）,提取ROI平均光谱。基于经典的过滤式特征选择算法拉普拉斯分数（Laplacian score）（无监督）和ReliefF（有监督）挑选建模关键变量,建立偏最小二乘回归模型（Partial least squares, PLS）。开发包含自定义选择层的一维卷积神经网络模型（One-dimensional convolutional neural networks, 1DCNN）。自定义层根据权重系数绝对值,可确定重要波长变量。Laplacian score-PLS模型对预测集中奶粉、三聚氰胺、香兰素和淀粉质量分数预测结果均方根误差分别为0.1110%、0.0570%、0.0349%和0.3481%。ReliefF-PLS模型对预测集中奶粉、三聚氰胺、香兰素和淀粉预测结果均方根误差分别为0.1998%、0.0540%、0.0455%和0.1823%。1DCNN模型对预测集中奶粉、三聚氰胺、香兰素和淀粉质量分数预测结果均方根误差分别为0.8561%、0.0911%、00644%和0.2942%。对Laplacian score、ReliefF和自定义选择层挑选出的前15个重要波长进行对比分析,不同特征选择方法挑选的特征波长子集有所区别,但都选择 1210、1474、1524、1680nm等附近波长。基于ReliefF-PLS模型的可视化结果表明了其良好的预测能力。     

A preliminary model to investigate culling and replacement policy in dairy herds

Replacement policy in dairy herds is not easy to determine when heifers are both home-reared and compete for land with the cows. The difficulty lies in evaluating quantitatively the advantages and disadvantages of low and high replacement rates. A computer simulation model was developed to examine the effects of different replacement rates on genetic improvement in yield per cow, milk output and farm profit at two levels of culling for reasons other than insufficient yield.The working hypothesis suggested by the results is that profitability falls as replacement rate increases because the number of cows able to be milked and the maturity of the herd over-ride the gains in milk yield per cow due to genetic improvement by yield culling and the greater use of a sire of high genetic merit.     

基于随机蛙跳和支持向量机的牛乳收购分级模型构建

蛋白质、脂肪含量和体细胞数量作为牛乳收购中的重要参考指标，决定了牛乳的品质和价格。为批量准确地对牛乳品质进行分级，提高乳企的生产效率，本研究以3216份荷斯坦牛牛乳样本为研究对象，应用中红外光谱技术实现对收购过程中4种不同品质牛乳的检测分级。利用一阶导数和一阶差分对光谱进行预处理，并结合竞争性自适应重加权算法（Competitive Adaptive Reweighted Sampling，CARS）和随机蛙跳算法（Shuffled Frog Leaping Algorithm，SFLA）筛选出能代表不同牛乳的有效特征变量，建立支持向量机（Support Vector Machine，SVM）模型。其中，利用网格搜索法（Grid Search，GS）、遗传算法（Genetic Algorithm，GA）和粒子群算法（Particle Swarm Optimization，PSO）对SVM模型的关键参数——惩罚参数c和核函数参数g进行优化。结果表明，SFLA算法总体上优于CARS算法，PSO优化SVM模型的效果最佳。一阶差分预处理后，利用SFLA算法筛选特征变量建立的PSO-SVM模型，训练集准确率、测试集准确率和曲线下面积（Area Under Curve，AUC）分别为97.8%、95.6%和0.96489。该模型具有较高的准确率，在牛乳产业中具有实际应用价值。     

利用可见―近红外光谱术无损检测牛奶中的三聚氰胺

初步探讨了利用可见―近红外光谱术检测牛奶中三聚氰胺的可行性及方法。通过往牛奶中掺入不同浓度三聚氰胺的方法,制备了165个样品,三聚氰胺浓度为0~1000ppm。利用光纤光谱仪采集样本的可见―近红外光谱,其光谱范围为350~1800nm。然后分别采用最小二乘法(PLS)、区间偏最小二乘法(IPLS)及联合区间偏最小二乘法(SIPLS)建立预测模型。比较实验结果表明:把光谱分为10个子区间,通过SIPLS方法,选出3个光谱子区间(4、7、9)联合建立的预测模型最优,其校正集和预测集得相关系数分别为0.9981和0.9946,校正集和预测集的均方根误差分别为0.1942和0.3299。因此,可见近红外光谱术结合联合区间偏最小二乘法能无损、快速的检测牛奶中的三聚氰胺。     

A short-term mechanistic model of forage and livestock in the semi-arid Succulent Karoo: 1. Description of the model and sensitivity analyses

This series of two papers describes a mechanistic model that simulates within years the productivity of vegetation and livestock on the communal semi-arid rangeland of the Succulent Karoo of South Africa. The model enables users to evaluate short-term management decisions on the production of milk and meat and to develop sets of equations and rules for long-term models designed to examine the effects of different strategies on the sustainability of the ecosystem.A soil moisture module partitions daily rainfall between runoff, infiltration and drainage and also simulates the loss of soil moisture by evaporation and transpiration. Forage production by different types of plant is modelled in relation to soil moisture and the present potential for growth. Three factors are assumed to influence the animal’s preference for a specific type of plant or part of a plant: relative abundance, ease of harvesting and digestibility. The model combines three mechanisms of food intake regulation: the rate at which the animal is able to eat forage, physical capacity of the digestive system, and, in young animals, their growth potential. Metabolisable energy intake is partitioned between maintenance, accretion/depletion of body protein and fat, conceptus growth and milk production. Reproductive and survival rates are simulated in relation to predicted liveweight and liveweight changes for the different age classes of livestock.     

基于电参数和支持向量回归的牛奶体细胞数定量检测

首先用碘酊染色显微镜体细胞计数法确定了100头不同患病程度的奶牛奶样的体细胞数,然后采用叉指电极在0.01～100 kHz频率范围对奶样进行了交流阻抗测试。在提出该测试系统等效电路的基础上用Zview软件获得了奶样电阻RS、电双层电容Cdl-T、Cdl-P等电参数,并分析了不同患病程度奶牛奶样的电参数特点。最后以镜检体细胞计数结果为标准,用电参数作为输入,建立了牛奶体细胞数的支持向量回归(SVR)预测模型,研究了体细胞数定量检测的可行性。结果表明,随着体细胞数的增加(即患病程度的增加),参数RS减小,而参数Cdl-T、Cdl-P呈非线性变化趋势。该定量预测模型对除N级以外所有奶样的体细胞数都有较高的预测精度,平均相对误差为29.40%。1级(隐性)、2级(较严重)和3级(严重)乳腺炎的检出正确率均达到100%。     

Simulation of the intake and partition of nutrients by the dairy cow: Part I — Management control in the dairy enterprise; philosophy and general model construction

The development of management control techniques for the dairy derd is discussed in this paper in relation to the rationale behind the construction of a detailed model of milk production in the individual cow. The movement towards a more flexible and individualised basis for predicting yield, which does not depend on historical records or standard energy systems, is highlighted.The overall concept and approach of the model are set out and the control aspects and structural framework of the program explained. Subsequent papers in the series will deal in more detail with the major sections of the program in order to illustrate how the mechanisms of the model actually operate and to provide results and an account of the validation procedure.     

基于Web的奶牛场管理信息系统

介绍基于JSP+SPrvlet+Javabean模式开发的、适合中型奶牛场使用的奶牛场管理信息系统,分为牛群管理、产乳管理、牛群繁殖、统计与分析、养牛场管理及系统管理功能6个方面,可以对奶牛及奶牛场各种技术数据进行分析与综合整理,是一种实用、经济和操作简便的奶牛场信息资料管理系统.     

Characteristics and economic viability of milk production in the smallholder farming systems in The Gambia

Domestic milk production has been for a long time hindered by many factors including lack of interest from decision makers, distorted economic policy and biotechnical constraints. For the last 20 years, many developing countries have been attempting to develop the domestic milk production sector. However, research on the basic realities and the viability status of enterprises within this sector remain largely unproved in many developing countries. This study focuses on the characteristic of smallholder milk producers in The Gambia. Data were collected from 90 smallholder farm households to characterise milk producers and evaluate the profitability and viability status of this activity. Based on current typology of farms and gross margin analyses at farm level, the study identified two resource-based types of smallholder farms. The current milk production system is surely viable. Constraints to increased productivity include lack of improved technology at farm level and weak institutional support. Despite the low viability status, it is shown that milk production generates reliable incomes, which could be a departure for most farmers to intensify farming systems, particularly in areas where no loan schemes exist for purchasing agricultural inputs.     

乳品企业技术创新能力评价指标体系与评价方法研究

在企业的技术创新过程中,技术创新能力的作用是至关重要的。为此,全面、系统地建立了乳品企业技术创新能力评价指标体系,并建立了评价模型,可以使乳品企业了解自身在技术创新方面的潜力和不足,对全面提高乳品技术创新能力具有一定的指导意义。